Revenue risk is the risk that realized revenue will be lower than expected or below a defined threshold. This post introduces a new series dedicated to revenue risk. I will write several pieces each month about this personal project, including the modeling choices, assumptions, results, and limitations.
The objective is not to replace a conventional revenue forecast. Instead, I want to build a framework that focuses specifically on downside outcomes: how likely a meaningful revenue impairment is, how much revenue is exposed when it occurs, and how severe the shortfall becomes afterward.
What is revenue risk?
Revenue risk is the risk that realized revenue will be lower than expected or below a certain threshold. A standard forecast typically estimates the most likely path for future revenue. A revenue risk model asks a different question: what is the probability and severity of a materially worse outcome?
How do I approach revenue risk?
My goal is to apply credit risk modeling techniques to revenue outcomes using historical data. I will adopt concepts such as Probability of Default (PD), Exposure at Default (EAD), and Loss Given Default (LGD), but reinterpret them for revenue impairment rather than debt default.
The terminology is inspired by established credit risk frameworks. The Basel Framework – IRB approach: risk components, for example, uses PD, LGD, and EAD as core credit risk components. My use of these concepts is an analogy rather than a regulatory credit risk model.
How is this different from existing revenue risk frameworks?
Existing approaches such as Revenue-at-Risk and Cash-Flow-at-Risk generally estimate the distribution or downside tail of future revenue or cash flow. Similar ideas are also used to quantify revenue exposed to specific disruptions.
The novel step explored here is the specific decomposition of revenue impairment into PD, EAD, and LGD: the probability of a material revenue decline, the counterfactual revenue exposed when it occurs, and the portion of that revenue that remains unrecovered over time. I do not assume that this decomposition is entirely unprecedented, but it is a different way of structuring the revenue downside problem.
What is revenue impairment?
When modeling default risk for debt instruments, a model developer has to decide what the definition of a default is. For example, it could mean the event of a missed principal repayment that is at least 90 days overdue.
I define “revenue impairment” as the event in which seasonally adjusted revenue falls 10% or more below its expected trend. It is similar to a default event in the sense that it represents a discrete downside event that can be assigned a probability and followed by a recovery path.
| Component | Credit risk interpretation | Revenue impairment implementation |
|---|---|---|
| Probability of Default (PD) | Probability that the borrower defaults within a defined horizon. | Probability that seasonally adjusted revenue falls >10% below expected trend in any quarter over the next 4 quarters. |
| Exposure at Default (EAD) | Amount of credit exposure outstanding when default occurs. | Counterfactual revenue exposed at the moment of impairment: the revenue that would have been expected in the impairment quarter absent the impairment. |
| Loss Given Default (LGD) | Fraction of EAD ultimately lost after recoveries following default. | Fraction of counterfactual revenue not recovered after impairment, measured from the revenue shortfall path until the end of a fixed horizon. |
How does PD work in a revenue risk model?
PD₁ is defined as the probability that seasonally adjusted revenue growth (quarter-on-quarter) falls more than 10% below its expected seasonal trend in any of the next four quarters. The historical pattern establishes normal seasonality, while the forecast horizon shows whether an impairment event breaches the threshold.
How does EAD work in a revenue risk model?
For a given revenue impairment event, EAD is the counterfactual revenue amount expected in the impairment quarter, assuming the impairment had not occurred. If 2025 Q2 expected revenue was $120m and actual revenue fell to $90m, then EAD = $120m. The $30m shortfall is the loss, not the exposure.
This distinction matters because EAD provides the revenue base against which the size of the impairment can be measured. It keeps the exposure separate from the realized shortfall.
How does LGD work in a revenue risk model?
LGD measures the cumulative severity of a revenue impairment once default has occurred. For each quarter in the fixed LGD horizon, the period-specific revenue shortfall is measured as a percentage of that period’s EAD. These shortfalls are then accumulated over time.
This makes LGD a path-dependent measure. A sharp one-quarter decline followed by a full recovery can produce a different cumulative loss than a smaller decline that persists for several quarters.
Why model revenue risk this way?
The purpose is to move beyond a regular revenue forecast and explicitly estimate downside risk. Separating probability, exposure, and loss severity creates a structured way to ask three different questions: how likely is the impairment, how much revenue is exposed when it happens, and how much of that exposure is ultimately lost over the recovery horizon?
This also creates a framework that can be tested historically. Each component can be estimated separately, validated against realized outcomes, and improved as more observations become available.
What comes next for the revenue risk model?
The next step is to estimate each revenue risk component using historical revenue data. I will explore how PD, EAD, and LGD can be modeled separately, how seasonality should be handled, and how the three components can eventually be combined into a single expected revenue loss measure.
The broader goal of this revenue risk series is to build the framework step by step, make the assumptions explicit, and test whether credit risk concepts can provide a useful way to quantify revenue downside.